Application of an Artificial Neural Network Model Based on Mineral Composition to the Prediction of Physical and Mechanical Properties

The physical and mechanical properties of soil are crucial in engineering construction, but conducting extensive experimental tests can be time-consuming, laborious, subject to uncertainties due the heterogeneity variations conditions. Soil is composed various minerals, mineral composition fundamental determinant soil. purpose this study establish a convenient reliable property prediction model based on composition. To achieve end, dataset comprising percentage content different minerals soil, as well soil’s properties, was collected. Using artificial neural network methods, models for liquid limit, plastic internal friction angle, cohesion were developed Each model’s performance evaluated through deviation analysis, with poor accuracy optimized using genetic algorithm. results demonstrate that accurately predict high applicability. This research provides method predicting majority data composition, which significant cost savings improving work efficiency projects.